Towards a general architecture for a co-learning of brain computer interfaces
Résumé
In this article we propose a software architecture for asynchronous BCIs based on co-learning, where both the system and the user jointly learn by providing feedback to one another. We propose the use of recent filtering techniques such as Riemann Geometry and ICA followed by multiple classifications, by both incremental supervised classifiers and minimally supervised classifiers. The classifier outputs are then combined adaptively according to the feedback using recursive neural networks.